3D human reconstruction from RGB images achieves decent results in good weather conditions but degrades dramatically in rough weather. To address this, mmWave radars have been employed to reconstruct 3D human joints and meshes in rough weather....
K-medoids clustering is a popular variant of k-means clustering, widely used in pattern recognition and machine learning. However, improper initialization can cause k-medoids clustering to get trapped in local optima. The INCKM algorithm was...
Stochastic optimisation algorithms are pivotal in machine learning, especially when dealing with large datasets. These algorithms work by processing only a subset of the available data at each step, which significantly reduces computational costs...
RadField3D is an open-source application based on Geant4 Monte-Carlo simulation, designed to generate three-dimensional radiation field datasets for dosimetry. This tool is particularly useful in the field of radiation-protection dosimetry for...
MegaSynth is a groundbreaking approach to 3D scene reconstruction that leverages synthesized data for training. At its core, MegaSynth is a procedurally generated 3D dataset comprising 700,000 scenes, significantly larger than previous datasets....
Ordinal Decision Trees are a machine learning approach specifically designed to handle ordinal classification tasks, where the labels exhibit a natural order. Unlike nominal classification, which treats all classes as equally distinct, ordinal...
The Certainty Ratio ($C_\rho$) is a novel metric introduced to assess the reliability of classifier predictions in machine learning. Traditional performance measures like accuracy and F-score often fail to account for the uncertainty inherent in...
Indirect Query Bayesian Optimization (IQBO) is a framework developed for a new class of Bayesian optimization problems where the integrated feedback is given via a conditional expectation of the unknown function to be optimized. The underlying...
Threshold Neurons are a novel artificial neuron model inspired by the threshold mechanisms and the excitation-inhibition balance observed in biological neurons. This model is designed to enhance the computational efficiency of on-device Deep Neural...
ManiBox is a novel approach in robotic grasping that addresses the challenge of spatial generalization in manipulation tasks. Traditional models often struggle with accurately positioning objects for grasping due to the extensive data requirements...
LeStrat-Net is a machine learning algorithm designed to enhance Monte Carlo simulations through a novel stratification approach. Traditional Monte Carlo methods divide the domain space of the integrand into regular intervals, but LeStrat-Net uses a...
Reinforcement Learning (RL) is a type of machine learning where agents learn to make decisions by receiving rewards or penalties for their actions. In neuroscience, RL has been used to model how the brain might implement learning and decision-making...
Distributionally Robust Learning (DRL) is a framework in machine learning that aims to ensure robust performance under distribution shifts. This is particularly important in scenarios where the data distribution at test time may differ from the...
The spiked Wigner model is a statistical model used to detect the presence of a signal in a noisy environment. It is particularly relevant in scenarios where the noise is non-Gaussian, and the signal is drawn from a Rademacher prior. The model...
Decentralized partially observable Markov decision processes (Dec-POMDPs) are a framework for modeling decision-making in multi-agent systems where each agent has limited information about the environment. The traditional approach to solving...
Large Language Models (LLMs) are a type of artificial intelligence model designed to understand and generate human-like text. These models are trained on vast amounts of text data, allowing them to learn the nuances of language, including grammar,...
Machine learning-based predictive models are a class of algorithms that use historical data to predict future outcomes. These models are built by training on datasets that contain input-output pairs, allowing the model to learn the relationship...
BirdVoxDetect is an open-source software tool powered by machine learning, designed to detect flight calls from songbirds in audio recordings. This tool is particularly useful for ornithologists and researchers studying bird migration patterns, as...
Machine learning is increasingly being used to analyze complex datasets in various scientific fields, including chemistry. In the context of heterogeneous catalyst data analysis, machine learning models can identify patterns and correlations in...
Neural Radiance Fields (NeRF) have revolutionized photorealistic novel view synthesis by enabling the generation of high-quality 3D representations from 2D images. However, NeRFs often suffer from artifacts known as 'floaters,' which degrade the...
Krony-PT is a compression technique for the GPT-2 language model, utilizing Kronecker Products to reduce the size of the model's MLP layers. This method systematically compresses the feed-forward layer matrices, resulting in smaller models with...
EEGNet is a deep learning architecture specifically designed for the analysis of electroencephalogram (EEG) data. It is particularly effective in tasks such as imagined speech detection, where it outperforms traditional machine learning classifiers...
Token Reduction using CLIP Metric (TRIM) is a novel approach designed to enhance the efficiency of Multimodal Large Language Models (MLLMs) by reducing the computational overhead associated with processing image tokens. Inspired by human attention...
Learning from Noisy Labels is a technique designed to train machine learning models effectively even when the training data contains incorrect or ambiguous labels. Traditional methods often require clean seed data or involve additional processing to...
Continuous Patient Monitoring with AI is an advanced platform developed by LookDeep Health for real-time analysis of video in hospital care settings. Leveraging computer vision, the platform provides insights into patient behavior and interactions...
PersonaMark is a novel personalized text watermarking scheme designed to protect the copyrights of Large Language Models (LLMs) and enhance accountability. With the rapid advancement of customized LLMs, safeguarding model copyright and ensuring data...
Domain Adaptive Object Detection (DAOD) is a technique aimed at generalizing an object detector trained on labeled source-domain data to a target domain without annotations. The core principle of DAOD is source-target feature alignment. Traditional...
Multi-View Incremental Learning (MVIL) is a framework designed to emulate the brain's ability to integrate sequentially arriving data views. It is inspired by bio-neurological processes and consists of two main modules: structured Hebbian plasticity...
Reefknot is a comprehensive benchmark designed to evaluate and mitigate relation hallucinations in multimodal large language models (MLLMs). It addresses the limitations of current benchmarks by providing a detailed evaluation framework and a novel...
Echo is a simulation framework designed to address the challenges of large-scale distributed training in machine learning. It focuses on tracing runtime training workloads, estimating collective communication, and accounting for computation slowdown...
SMOSE is a novel method for training interpretable controllers in reinforcement learning for continuous control tasks. It uses a Sparse Mixture of Shallow Experts architecture to combine interpretable decision-makers and a router for task...
SEG-SAM is a unified medical image segmentation model that enhances performance by incorporating semantic medical knowledge. It addresses the challenges of transferring the Segment Anything Model (SAM) to the medical domain, where images often have...
Monocular facial appearance capture is a technique used to reconstruct the appearance properties of human faces from a single camera setup, typically in an unconstrained environment. This method is particularly useful for applications where...
Deep Learning for Hydroelectric Optimization focuses on generating long-term river discharge scenarios using ensemble forecasts from global circulation models. Hydroelectric power generation is a critical component of the global energy matrix,...
The Advantage-based Optimization Method for Reinforcement Learning addresses the challenges of large, high-dimensional action spaces in real-world scenarios. Traditional value-based reinforcement learning algorithms often struggle with convergence...
AutoSciLab is an innovative machine learning framework designed to automate scientific experiments, effectively acting as a surrogate researcher. This framework is particularly useful in high-dimensional spaces where traditional experimental design...
Large Language Models (LLMs) have gained prominence for their ability to handle complex data and extract meaningful insights. This study investigates the effectiveness of LLMs in time series data analysis, a critical task across domains like...
The Lagrangian Index Policy (LIP) is a heuristic approach used in the field of reinforcement learning, particularly for solving restless multi-armed bandit problems. These problems involve making decisions over time to maximize rewards, where each...
Bilevel optimization is a sophisticated mathematical framework used to solve problems where one optimization problem is nested within another. In the context of imbalanced data classification, this framework is particularly useful as it allows for...
Optimal control problems (OCPs) are mathematical problems that involve finding a control policy for a dynamical system to optimize a certain performance criterion. These problems are central to many applications in engineering and science, where the...